Unveiling and Quantifying Facebook Exploitation of Sensitive Personal Data for Advertising Purposes
summary
Analysis uses data from 45,500 real Facebook users yielding ~5 million interest assignments (126,000 unique), scores each interest with semantic-similarity to GDPR-sensitive keywords, has 12 panelists label a 4,400-candidate subset and validates 2,092 sensitive interests, queries Facebook Ads Manager API for audience sizes, finds 73% of EU Facebook users (≈ 40% of EU citizens) tagged with potentially sensitive interests, runs ad experiments that reach 26,000 users for €35 (implying ~€0.015 per user deanonymization cost), and ships a browser extension to show assigned sensitive interests.
AI-written summary · about
chapters (10)
- 0:00Intro
- 0:37Motivation
- 4:05Classification
- 11:42Reasons of Interest
- 12:31Ad Targeting
- 13:21Hate Campaigns
- 15:20Impact
- 16:53Conclusion
- 17:26Results
- 18:57Questions
related talks
51mDiscover how your generic data makes you unique among millions of users
01h 24mfrom “old” filter bubbles to political microtargeting.
38mFireside Chat: Personalisation and Privacy - At Odds or Not?
01h 07mThe great data race ...
01h 07mPolitical behavioural targeting
23mKeynote: Digital Psychometrics and it's Future Effects on Technology
32mPlatform-supported Auditing Of Social Media Algorithms For Public Interest
57mData Protection & Digital Marketing – An interactive networking session